um: switch to regset API and depend on XSTATE

The PTRACE_GETREGSET API has now existed since Linux 2.6.33. The XSAVE
CPU feature should also be sufficiently common to be able to rely on it.

With this, define our internal FP state to be the hosts XSAVE data. Add
discovery for the hosts XSAVE size and place the FP registers at the end
of task_struct so that we can adjust the size at runtime.

Next we can implement the regset API on top and update the signal
handling as well as ptrace APIs to use them. Also switch coredump
creation to use the regset API and finally set HAVE_ARCH_TRACEHOOK.

This considerably improves the signal frames. Previously they might not
have contained all the registers (i386) and also did not have the
sizes and magic values set to the correct values to permit userspace to
decode the frame.

As a side effect, this will permit UML to run on hosts with newer CPU
extensions (such as AMX) that need even more register state.

Signed-off-by: Benjamin Berg <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Johannes Berg <[email protected]>

Author: benzea

arch/um/Kconfig

@@ -5,6 +5,7 @@ menu "UML-specific options"
 config UML
 	bool
 	default y
+	select ARCH_WANTS_DYNAMIC_TASK_STRUCT
 	select ARCH_HAS_CPU_FINALIZE_INIT
 	select ARCH_HAS_FORTIFY_SOURCE
 	select ARCH_HAS_GCOV_PROFILE_ALL
@@ -32,6 +33,7 @@ config UML
 	select HAVE_ARCH_VMAP_STACK
 	select HAVE_RUST
 	select ARCH_HAS_UBSAN
+	select HAVE_ARCH_TRACEHOOK
 
 config MMU
 	bool

arch/um/include/asm/processor-generic.h

@@ -20,7 +20,6 @@ struct task_struct;
 struct mm_struct;
 
 struct thread_struct {
-	struct pt_regs regs;
 	struct pt_regs *segv_regs;
 	struct task_struct *prev_sched;
 	struct arch_thread arch;
@@ -31,6 +30,9 @@ struct thread_struct {
 			void *arg;
 		} thread;
 	} request;
+
+	/* Contains variable sized FP registers */
+	struct pt_regs regs;
 };
 
 #define INIT_THREAD \

arch/um/kernel/process.c

@@ -187,6 +187,13 @@ void initial_thread_cb(void (*proc)(void *), void *arg)
 	kmalloc_ok = save_kmalloc_ok;
 }
 
+int arch_dup_task_struct(struct task_struct *dst,
+			 struct task_struct *src)
+{
+	memcpy(dst, src, arch_task_struct_size);
+	return 0;
+}
+
 void um_idle_sleep(void)
 {
 	if (time_travel_mode != TT_MODE_OFF)
@@ -287,18 +294,3 @@ unsigned long __get_wchan(struct task_struct *p)
 
 	return 0;
 }
-
-int elf_core_copy_task_fpregs(struct task_struct *t, elf_fpregset_t *fpu)
-{
-#ifdef CONFIG_X86_32
-	extern int have_fpx_regs;
-
-	/* FIXME: A plain copy does not work on i386 with have_fpx_regs */
-	if (have_fpx_regs)
-		return 0;
-#endif
-	memcpy(fpu, &t->thread.regs.regs.fp, sizeof(*fpu));
-
-	return 1;
-}
-

arch/um/kernel/um_arch.c

@@ -402,6 +402,8 @@ int __init linux_main(int argc, char **argv, char **envp)
 		os_info("Kernel virtual memory size shrunk to %lu bytes\n",
 			virtmem_size);
 
+	arch_task_struct_size = sizeof(struct task_struct) + host_fp_size;
+
 	os_flush_stdout();
 
 	return start_uml();

arch/um/os-Linux/registers.c

@@ -10,11 +10,12 @@
 #include <sysdep/ptrace.h>
 #include <sysdep/ptrace_user.h>
 #include <registers.h>
+#include <stdlib.h>
 
 /* This is set once at boot time and not changed thereafter */
 
 static unsigned long exec_regs[MAX_REG_NR];
-static unsigned long exec_fp_regs[FP_SIZE];
+static unsigned long *exec_fp_regs;
 
 int init_pid_registers(int pid)
 {
@@ -24,7 +25,11 @@ int init_pid_registers(int pid)
 	if (err < 0)
 		return -errno;
 
-	arch_init_registers(pid);
+	err = arch_init_registers(pid);
+	if (err < 0)
+		return err;
+
+	exec_fp_regs = malloc(host_fp_size);
 	get_fp_registers(pid, exec_fp_regs);
 	return 0;
 }
@@ -34,5 +39,5 @@ void get_safe_registers(unsigned long *regs, unsigned long *fp_regs)
 	memcpy(regs, exec_regs, sizeof(exec_regs));
 
 	if (fp_regs)
-		memcpy(fp_regs, exec_fp_regs, sizeof(exec_fp_regs));
+		memcpy(fp_regs, exec_fp_regs, host_fp_size);
 }

arch/x86/um/Makefile

@@ -10,7 +10,7 @@ else
 endif
 
 obj-y = bugs_$(BITS).o delay.o fault.o \
-	ptrace_$(BITS).o ptrace_user.o setjmp_$(BITS).o signal.o \
+	ptrace.o ptrace_$(BITS).o ptrace_user.o setjmp_$(BITS).o signal.o \
 	stub_segv.o \
 	sys_call_table_$(BITS).o sysrq_$(BITS).o tls_$(BITS).o \
 	mem_$(BITS).o subarch.o os-Linux/

arch/x86/um/asm/elf.h

@@ -8,6 +8,8 @@
 #include <asm/user.h>
 #include <skas.h>
 
+#define CORE_DUMP_USE_REGSET
+
 #ifdef CONFIG_X86_32
 
 #define R_386_NONE	0

arch/x86/um/asm/ptrace.h

@@ -2,6 +2,16 @@
 #ifndef __UM_X86_PTRACE_H
 #define __UM_X86_PTRACE_H
 
+/* This is here because signal.c needs the REGSET_FP_LEGACY definition */
+enum {
+	REGSET_GENERAL,
+#ifdef CONFIG_X86_32
+	REGSET_FP_LEGACY,
+#endif
+	REGSET_FP,
+	REGSET_XSTATE,
+};
+
 #include <linux/compiler.h>
 #ifndef CONFIG_X86_32
 #define __FRAME_OFFSETS /* Needed to get the R* macros */

arch/x86/um/os-Linux/registers.c

@@ -16,133 +16,58 @@
 #include <asm/sigcontext.h>
 #include <linux/elf.h>
 #include <registers.h>
+#include <sys/mman.h>
 
-static int have_xstate_support;
+unsigned long host_fp_size;
 
-static int save_i387_registers(int pid, unsigned long *fp_regs)
-{
-	if (ptrace(PTRACE_GETFPREGS, pid, 0, fp_regs) < 0)
-		return -errno;
-	return 0;
-}
-
-static int save_fp_registers(int pid, unsigned long *fp_regs)
+int get_fp_registers(int pid, unsigned long *regs)
 {
-#ifdef PTRACE_GETREGSET
-	struct iovec iov;
+	struct iovec iov = {
+		.iov_base = regs,
+		.iov_len = host_fp_size,
+	};
 
-	if (have_xstate_support) {
-		iov.iov_base = fp_regs;
-		iov.iov_len = FP_SIZE * sizeof(unsigned long);
-		if (ptrace(PTRACE_GETREGSET, pid, NT_X86_XSTATE, &iov) < 0)
-			return -errno;
-		return 0;
-	} else
-#endif
-		return save_i387_registers(pid, fp_regs);
-}
-
-static int restore_i387_registers(int pid, unsigned long *fp_regs)
-{
-	if (ptrace(PTRACE_SETFPREGS, pid, 0, fp_regs) < 0)
+	if (ptrace(PTRACE_GETREGSET, pid, NT_X86_XSTATE, &iov) < 0)
 		return -errno;
 	return 0;
 }
 
-static int restore_fp_registers(int pid, unsigned long *fp_regs)
-{
-#ifdef PTRACE_SETREGSET
-	struct iovec iov;
-	if (have_xstate_support) {
-		iov.iov_base = fp_regs;
-		iov.iov_len = FP_SIZE * sizeof(unsigned long);
-		if (ptrace(PTRACE_SETREGSET, pid, NT_X86_XSTATE, &iov) < 0)
-			return -errno;
-		return 0;
-	} else
-#endif
-		return restore_i387_registers(pid, fp_regs);
-}
-
-#ifdef __i386__
-int have_fpx_regs = 1;
-static int save_fpx_registers(int pid, unsigned long *fp_regs)
+int put_fp_registers(int pid, unsigned long *regs)
 {
-	if (ptrace(PTRACE_GETFPXREGS, pid, 0, fp_regs) < 0)
-		return -errno;
-	return 0;
-}
+	struct iovec iov = {
+		.iov_base = regs,
+		.iov_len = host_fp_size,
+	};
 
-static int restore_fpx_registers(int pid, unsigned long *fp_regs)
-{
-	if (ptrace(PTRACE_SETFPXREGS, pid, 0, fp_regs) < 0)
+	if (ptrace(PTRACE_SETREGSET, pid, NT_X86_XSTATE, &iov) < 0)
 		return -errno;
 	return 0;
 }
 
-int get_fp_registers(int pid, unsigned long *regs)
-{
-	if (have_fpx_regs)
-		return save_fpx_registers(pid, regs);
-	else
-		return save_fp_registers(pid, regs);
-}
-
-int put_fp_registers(int pid, unsigned long *regs)
-{
-	if (have_fpx_regs)
-		return restore_fpx_registers(pid, regs);
-	else
-		return restore_fp_registers(pid, regs);
-}
-
-void arch_init_registers(int pid)
-{
-	struct user_fpxregs_struct fpx_regs;
-	int err;
-
-	err = ptrace(PTRACE_GETFPXREGS, pid, 0, &fpx_regs);
-	if (!err)
-		return;
-
-	if (errno != EIO)
-		panic("check_ptrace : PTRACE_GETFPXREGS failed, errno = %d",
-		      errno);
-
-	have_fpx_regs = 0;
-}
-#else
-
-int get_fp_registers(int pid, unsigned long *regs)
+int arch_init_registers(int pid)
 {
-	return save_fp_registers(pid, regs);
+	struct iovec iov = {
+		/* Just use plenty of space, it does not cost us anything */
+		.iov_len = 2 * 1024 * 1024,
+	};
+	int ret;
+
+	iov.iov_base = mmap(NULL, iov.iov_len, PROT_WRITE | PROT_READ,
+			    MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+	if (iov.iov_base == MAP_FAILED)
+		return -ENOMEM;
+
+	/* GDB has x86_xsave_length, which uses x86_cpuid_count */
+	ret = ptrace(PTRACE_GETREGSET, pid, NT_X86_XSTATE, &iov);
+	if (ret)
+		ret = -errno;
+	munmap(iov.iov_base, 2 * 1024 * 1024);
+
+	host_fp_size = iov.iov_len;
+
+	return ret;
 }
 
-int put_fp_registers(int pid, unsigned long *regs)
-{
-	return restore_fp_registers(pid, regs);
-}
-
-void arch_init_registers(int pid)
-{
-#ifdef PTRACE_GETREGSET
-	void * fp_regs;
-	struct iovec iov;
-
-	fp_regs = malloc(FP_SIZE * sizeof(unsigned long));
-	if(fp_regs == NULL)
-		return;
-
-	iov.iov_base = fp_regs;
-	iov.iov_len = FP_SIZE * sizeof(unsigned long);
-	if (ptrace(PTRACE_GETREGSET, pid, NT_X86_XSTATE, &iov) == 0)
-		have_xstate_support = 1;
-
-	free(fp_regs);
-#endif
-}
-#endif
-
 unsigned long get_thread_reg(int reg, jmp_buf *buf)
 {
 	switch (reg) {